THE BIRDS OF AUSTRALIA. 
as approaching, in many respects, the Falconinae again. The difficulty of 
being tied down to such a system is obvious : each form must fall into 
these five, and as each subfamily again was only allowed five genera and 
each genus five subgenera, the limits of the Falconidse was one hundred 
and twenty-five subgenera. I have not studied the system to see if each 
subgenus was limited to five species, but if so the species were also absolutely 
limited. To this drawback was contrasted the fact that many generic and 
subgeneric forms to fit into the circle had not been discovered, so that if 
a genus or subgenus were becoming overcrowded, the species could be drafted 
into another circle. Such an ingenious arrangement seems quite unnatural 
and met with violent opposition which soon killed it, yet the workers who 
were attracted by it were the most accurate and persevering of their day. 
Kaup then divided the subfamily Falconinse into the genera Hierax, 
Tinnunculus, Harpagus, Falco and leracidea. The first, thirds and fifth 
proved indivisible, but the second and fourth provided a full quinary 
complement of subgenera, viz., Polihierax, Eryihropus, PoecilorniSy Tichornis 
and Tinnunculus in the first case, and Aesalon, Hypotriorchis, Gennaia, 
Falco and Hierofalco in the other. 
In a similar manner the other subfamilies and genera were split up. 
The majority of these divisions were first proposed by Kaup^ and if Kaup 
did nothing else he made other workers look at the birds to see whether 
the differences he recorded existed or not. That is the one advantage of 
splitting, and it must be as applicable to genera as species. It is a remark- 
able fact that almost every worker who devotes himself to a group becomes 
a genus-splitter, and only superficial observers continue genus-lumping. 
We have not yet studied any group so fully that we can dogmatise what 
is a genus or not : a genus must be a natural association of species ; 
if the association of species be proved unnatural on any grounds, then the 
best purpose will be served by disintegration. Thus, Falco was subdivided 
by Kaup, as noted above, and it will be observed that he admitted 
Tinnunculus { — CercJineis of this time) and Erythropus. Genus- lumpers 
would include these latter again in Falco, urging that the last named is a 
connecting link. I will discuss this item more fully later on, bht I would 
here state that CercJineis seems absolutely differentiated, and, if included, no 
line can be drawn with accuracy. The usage of a different name is a slight 
matter if better definition is thereby secured. If it be convenient to handle 
hundreds of trinomials, why should convenience stick at a few different 
generic names ? Kaup’s subdivisions have been more or less admitted 
since his day and they have been transferred backwards and forwards in 
endeavours to provide natural systems. 
hi 
4 
